← ArchivePaper2014

Practice and Theory of Blendshape Facial Models

J P Lewis, Ken Anjyo, Taehyun Rhee, Mengjie Zhang, Frederic Pighin, Zhigang Deng

EurographicsAcademic325 cites3 descendantsFacial

State-of-the-art report on blendshape facial models covering theory, practice, workflow, and mathematical foundations used in VFX and games.

Open the paper ↗Trace in the lineage graph

Abstract

This state of the art report surveys the practice and theory of blendshapes, a simple linear model of facial expression that represents a face as a linear combination of sculpted expression targets and is the prevalent approach to realistic facial animation in film and commercial packages. The report covers blendshape terminology and history, the linear algebra of whole-face and delta formulations, combination and intermediate shapes, methods for constructing models, and animation techniques including keyframing, performance-driven capture, expression cloning, and direct manipulation. It contrasts the interpretable semantic blendshape basis with orthogonal PCA models and frames facial animation as a high-dimensional scattered interpolation problem. The authors show that despite the simplicity of the approach, several open problems remain associated with this fundamental technique.

How to read this

Category: Survey / state-of-the-art report on blendshape facial models
Contributions: A survey of blendshape terminology, history, and the linear algebra of whole-face and delta formulations, including combination and intermediate shapes
A review of model-construction methods and animation techniques (keyframing, performance-driven capture, expression cloning, direct manipulation)
A framing of facial animation as high-dimensional scattered interpolation, contrasting interpretable semantic blendshapes with orthogonal PCA models, and a list of open problems
Context: Consolidates the blendshape literature used across VFX and games, building on direct-manipulation work (Lewis and Anjyo's 'Direct Manipulation Blendshapes') and the broader linear-facial-model lineage.Builds on: Direct Manipulation Blendshapes
Correctness: As a state-of-the-art report it organizes and frames existing methods rather than presenting new validated results; treat it as an authoritative map and a source of open problems, not as empirical evidence for any single technique.
Clarity: Highly accessible and pedagogical; a first pass gives the conceptual landscape, and a second pass pays off for the delta/whole-face linear algebra and the scattered-interpolation framing.
How to read it: Use it as your entry point and reference for the whole topic; first pass for vocabulary and the map of methods, then return to specific sections (linear algebra, construction, manipulation) as needed for the technique you care about.

Builds on

Direct Manipulation Blendshapes 2010

Built upon by

Sketch-Based Controllers for Blendshape Facial Animation 2015

Related work

Direct Manipulation Blendshapes 2010 / TOG
Animating Facial Expressions 1981 / SIGGRAPH
Transferring the Rig and Animations from a Character to Different Face Models 2008 / CGF
Reusable Facial Rigging and Animation: Create Once, Use Many 2007 / PhD Thesis

Keywords

blendshapes blendshape targets facial animation delta blendshape combination blendshapes semantic parameterization principal component analysis morphable model performance-driven animation expression cloning parallel parameterization facs direct manipulation radial basis function high dimensional interpolation keyframe animation facial rigging linear blending

This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →